Evaluating the performance of nonergodic ground motion models in the ridgecrest area

The current state of the practice in performance-based earthquake engineering relies on using ergodic ground motion models (GMMs), which assume that the ground motion variability observed in a global database is the same as the variability in ground motion at a single site-source combination. However, as empirical ground motion databases have grown, it has become clear that there are significant systematic differences, which depend on repeatable effects for a combination of sites and seismic sources in a particular region. These systematic differences do not support the ergodic assumption, which is prompting the transition from ergodic to nonergodic GMMs. In this study, we use the Ridgecrest ground motion database, which has 20,000 + ground motion recordings, to evaluate the performance of ergodic and nonergodic GMMs. The large number of ground motions in the Ridgecrest database allows us to quantify repeatable effects for a relatively small region, considering earthquakes with a range of magnitudes, which has been seldom attempted in previous efforts. In developing the nonergodic GMMs, we propose a novel approach based on computer graphics to quantify the cell-specific attenuation terms to constrain the path effects. We use the developed ergodic and nonergodic GMMs to evaluate their performance in earthquake scenarios that occurred in the Ridgecrest area by creating GMM-based maps of ground shaking (MGSs) and comparing them with actual MGSs from recorded ground motions. We use the results from MGSs to assess the assumption in nonergodic GMMs, namely that repeatable effects observed in small magnitude earthquakes are correlated with repeatable effects observed in large magnitude earthquakes. Our results show that the nonergodic GMMs provide information on the spatial variation of repeatable effects induced by complex physical processes; they have a lower aleatory variability, consistent with previous studies; they have a better performance for the considered earthquake scenarios in the Ridgecrest area; and the repeatable effects quantified in small magnitude earthquakes improve the estimation of intensity measures (IMs) at larger magnitudes.